I sourced 100 LT3080 T0-220 Parts, will list on ebay.

[minime72706]

NOTE: I'VE NOTICED THAT THERE IS A UK SELLER OF THESE; I'D PROBABLY BE MORE CONVENIENT FOR THOSE IN USA/CANADA.

Hey y'all!

I've been thinking about building an adjustable, CV/CC, computer-controlled, DC power supply for a long time and I've been a viewer of EEVblog off and on. I came across the power supply series last month and took an interest in it. I was most pleased to be introduced to a better alternative to the LM317, the LT3080. I sourced 100 parts in the TO-220 package directly from Linear Technologies because the price break at 100 and shipping costs made it impractical to buy something like only 10. They seem somewhat hard to get and surprisingly there's no one on eBay selling them, so I bought way more than I needed to justify (in my mind) buying them at all. I had hoped that some people would be happy to have a different route to purchase them as well, so hopefully they don't sit around!

I'll likely sell the LT3080 in quantities of five (5) for something like $30 shipped. They're $3.14 a pop in quantities under 100 and shipping (for me in MA, USA) was about $11, so it seems fairly reasonable to me. I may drop the price and then charge actual postage to make it a bit more fair, but that's somewhat of a pain for me. After seeing Dave's video about anti-static and static-dissipative bags, I'll certainly use real 3M metal-in anti-static bags! I've not received the parts yet as I've only ordered them a few days ago (and they appear to be back-ordered); I'm posting preemptively to gauge interest! My eBay seller name is the same as my username on this forum, but again, you'll find nothing right now! I'll update when I receive parts and are ready to ship. I'll probably keep 20-25 for myself and put the rest up for sale.

Thanks!

[free_electron]

Please do explain why you think the LT3080 is better than an LM317....

[minime72706]

Why don't you watch his video series on his power supply design? This is not MY idea. (when searching the episode list, beware, he changes the title 3 times in the series of like 8-9 episodes)

Since it uses a current source and resistor as a reference, one can force a voltage into the control pin and achieve output voltages closely approaching 0V.
With the LM317, a 1.25V voltage reference is used and any voltage you put on the ADJ pin is in SERIES with that 1.25V. In order to operate down close to 0V you would normally need a negative voltage.
The LT3080 also boasts a much lower drop-out voltage.
Furthermore, it is rather simple to safely put the LT3080 in parallel with more of its kind to get a higher output current. With some resistive ballasting you can design your circuit so they share the power load rather evenly.

You can generally design a circuit using an LM317 to do what you want, but the LT3080 simplifies things in many ways. I might even be less noisy and more stable, but it's not important enough for me to compare datasheets.

The use of an ellipse (....) signifies impatience; there's no need for that.

[free_electron]

MY question was based on this :

I've been thinking about building an adjustable, CV/CC, computer-controlled, DC power supply for a long time

here is the problems with the LT3080 for that given setup:

1) no current control... you'll have to create another regulation loop that will have to fight the LT3080 regulation

2) max 1.1 ampere.. the LM317 does 1.5

3) low drop.. who cares.. a lab power supply burns off 90% of the energy anyway .. you start with 20 or 30 volt at the input to make 5 volt at the ouptut. low-drop is a non issue in this case

4) controllable to 0 volt. so is an lm317 if you use a negative reference voltage.

5) Neither the lt3080 or the lm317 are advisable for lab power supply usage. Don't get me wrong, they are great regulators , but they are intended to be put in a system where a set of resistors aligns them to make a specific voltage and the STAY there for the rest of their useful life. You do not slap a potentiometer on them and build a 'lab power supply with them
they are no where near robust enough , too many limitations , no protection , no controllable current limit no nothing. They are intended for a finished product. not a lab instrument.

and then there is this whole 'computer control' stuff. we are going to spend 50$ in parts to slap on a fancy display and computer control but it is too cumbersome and 'expensive' to buld a decent regulator section. ( which would cost like 10$ ). Of course hammering software on a pic and pc is easier than making a good analog control loop. This is my gripe with these kinds of projects.
Note : i am NOT attacking you. It's just that the internet is full of total crap designs made by people who have an affliction for writing code but have no clue how to properly use an opamp and transistor. The end result is always pure garbage. If you are really serious about making a good DC power supply , that is computer controllable , start with building a solid regulator system and provide a properly protected power section. and THEN slap on a pic if you want to.

[BravoV]

here is the problems with the LT3080 for that given setup .... <snip>...

... <snip>...

5) Neither the lt3080 or the lm317 are advisable for lab power supply usage.

.... <snip>...

Note : i am NOT attacking you. It's just that the internet is full of total crap designs made by people who have an affliction for writing code but have no clue how to properly use an opamp and transistor. The end result is always pure garbage. If you are really serious about making a good DC power supply , that is computer controllable , start with building a solid regulator system and provide a properly protected power section. and THEN slap on a pic if you want to.

Assuming you've watched those videos, I guess you're politely saying that Dave's design is also total crap since he is using LT3080 for his upcoming psu that he is going to sell.

Just correct me if I'm wrong. 

[jsoderquist]

Ellipse?

"You keep using that word... I do not think it means what you think it means..." --Inigo Montoya

[free_electron]

I have not watched those videos, and that is NOT what i am saying.

Dave is using the 3080 to make a usb-powered , portable doohikey to use in the field as a stop-gap for not having to lug around a real lab power supply.
If that is your intention the LT3080 will do just fine. IF i needed to design a usb powered adjustable supply i might do the same.

Then again, i wouldn't be designing such a thing as i have zero interest in that. It is bad to use a computer usb port as power output. They are not intended for that purpose and i do not want to fry a 2500$ laptop. And it doesn't work on all usb ports. I can see a scenario will people with a laptop that has 4 usb ports will hook up a usb hub and connect 5 of these little power supplies and crank away... it'll all end up in tears .

I'll lug around a real power supply. Actually one may be better off by buying a laptop power brick that gives 19 volts at 4 ampere and combine that with a circuit board with power transistor and regulator and use that as a low-weight transportable power supply. those power briks are robust, pack a lot of power , are light. add a little circuit board to that, two potmeters , 2 opamps a darlington power stage style TIP142 add a pic with an lcd display to show set voltage , current and real voltage and current. and you will have something that works way better, has more power , and doesn't need to be hooked up to a computer.

[minime72706]

1. Sorry, I meant "ellipsis". I goofed. I'm a bit out of it (mono/glandular fever).

2. The USB power supply series, which I haven't watched yet, is a different design.

3. I hate writing code; you have no idea how much I hate writing code. More experience in doing so, though painful, is still useful to me.

My design is still evolving, but I'm looking at a 9VAC/18VAC  or (10/20) transformer to switch between two ranges to save burning some energy in the linear regulation section.
I've also considered a switching pre-regulator, but my experience in the past with the MC34063 is not stellar. (24VAC -> 3.3VDC = kaboom in my design)
0-20VDC@2A adjustable section (two LT3080s in parallel is what's on my mind right now) and probably a 5V@1A fixed-output (I will use a SEPARATE regulator for the control circuitry, I don't want a short on the 5V output to shut down the regulation circuitry!)

I have a fairly good BK Precision triple-output power supply. This is mostly for the heck of it, but if I end up with a semi-portable adjustable power supply, that's a bonus. A design LIKE IT might be useful at work to reduce the number of expensive DC supplies required when testing a chip (I make integrated circuits).

PS - If anyone can find a ~40-50VA transformer with a 120VAC primary and something like a 6/12/24VAC secondary (three taps), I'd be RATHER grateful as it would improve my power dissipation. I've seen them before on surplus sites and on ebay, but I've otherwise come out empty-handed.

EDIT - Heh, I just realized that my advertisement has turned into a debate about power supply design. I hope the thread doesn't get moved! 

EDIT 2 - Also, the computer control is meant to give one the kind of functionality that a high-end Agilent supply gives you - automated testing. It will also have some knobs to control the supply in lieu of a computer.
I haven't really given it much thought yet, but has anyone a good idea for safely switching between the 9VAC and 18VAC secondaries while the PSU is in operation? I wouldn't want a spike on the output when I cross the threshold.

[madires]

Hi!

Then again, i wouldn't be designing such a thing as i have zero interest in that. It is bad to use a computer usb port as power output. They are not intended for that purpose and i do not want to fry a 2500$ laptop. And it doesn't work on all usb ports. I can see a scenario will people with a laptop that has 4 usb ports will hook up a usb hub and connect 5 of these little power supplies and crank away... it'll all end up in tears .

LOL! Using a PC or laptop (powered by the PSU) as power source requires some noise/EMI filtering. If you power an audio amp by the PCs PSU you'll be able to hear changing CPU loads. After adding some caps and chokes it's mostly gone. For cheap PSUs it's even more worse.

I'll lug around a real power supply. Actually one may be better off by buying a laptop power brick that gives 19 volts at 4 ampere and combine that with a circuit board with power transistor and regulator and use that as a low-weight transportable power supply.

Do you got a shortwave receiver? Put the laptop PSU and the shortware receiver in the same room and have some fun.

Cheers
 madires

[minime72706]

LOL! Using a PC or laptop (powered by the PSU) as power source requires some noise/EMI filtering. If you power an audio amp by the PCs PSU you'll be able to hear changing CPU loads. After adding some caps and chokes it's mostly gone. For cheap PSUs it's even more worse.

Heh, that sound of the changing CPU (and GPU) load drives me INSANE. You can hear it when you move the mouse...
It's much less apparent if you use the rear port of a desktop computer (direct from the sound card) instead of the headphone port that might be located on the front

[nctnico]

I'll lug around a real power supply. Actually one may be better off by buying a laptop power brick that gives 19 volts at 4 ampere and combine that with a circuit board with power transistor and regulator and use that as a low-weight transportable power

If you can spend the money, get a lightweight PSU like these:
http://www.delta-elektronika.nl/en/products/es150-series.html

[minime72706]

Those supplies are neat, but the purpose of this is mostly just to do a project with a practical purpose (one that I'll be apt to see to the end) and possibly to integrate the design into a test fixture at work.

EDIT: By the way, unless I'm mistaken, the PSU I have is the BK Precision 1672.

[Bored@Work]

The Lascar PSU 130 is nice, except that it doesn't have an adjustable current limit.

[free_electron]

I've also considered a switching pre-regulator, but my experience in the past with the MC34063 is not stellar. (24VAC -> 3.3VDC = kaboom in my design)

MC34063   Motorola Crap... that thing is one of the worst switchers out there...  I really don;t get why people seem to be stuck at LM317's , NE555's , LM324's and MC34063's...
Almost anywhere you look on the internet, hobbyists are endlessly recycling that old tripe... All the while when there are hundreds if not thousands way better circuits out there.

0-20VDC@2A adjustable section (two LT3080s in parallel is what's on my mind right now)

Why not go to a real power transistor like a TIP142 ? my gut feelins still says the LT3080 is not intended for this kind of lab supply work. Reverse feed it and it will most certainly fry...

Also, the computer control is meant to give one the kind of functionality that a high-end Agilent supply gives you - automated testing. It will also have some knobs to control the supply in lieu of a computer.

better make sure you have good readback capability then.. and that the supply is stable under impulse loads.

I haven't really given it much thought yet, but has anyone a good idea for safely switching between the 9VAC and 18VAC secondaries while the PSU is in operation? I wouldn't want a spike on the output when I cross the threshold.

that's a matter of having good line regulation. i would switch over using SCR's or mosfets. you can actually make a chopping system where you seamlessly go from one to another.
BUt it's going to be hard getting the performance of , let's say a 3632 ...

[minime72706]

I'm not averse to using a switcher, I was just making a joke of my past misfortunes with that chip. (My senior design partner almost lost his eye to a fragmenting device)

Dave tested the LT3080's reverse protection, and as long as you don't short the input to ground and feed a voltage in backwards, it survives.

Elaborate on what you're questioning with the PC-control, please.

I'll look into your advice regarding my secondary-switching problem, thanks for that.

[T4P]

The best regulator design is always a opamp + big fat chunky transistor
For switching the voltage taps use relays and some protection diodes to prevent the inductance from killing the circuits
Or for switching a not so complex regulation circuit and well tuned filters make the best out of a power supply

[minime72706]

I have trouble arguing with that, heh.
There's something to be said about tuning for stability and thermal protection, though.

[T4P]

I have trouble arguing with that, heh.
There's something to be said about tuning for stability and thermal protection, though.

the classic circuit still is far more stable

[tom66]

My (purchased, not made) bench power supply uses SCRs to switch taps. It's quite a clever solution; the gates are biased such that when the output goes over a certain setpoint, it will switch taps. Because it uses SCRs (after the bridge) it has automatic delay (of one AC cycle), so won't jitter and it is a cheap, reliable solution.

I would suggest one low voltage (3-4V) tap, so you can handle shorts easily, and then two middle of the range taps, so that total power dissipation is limited to around 10W.

[minime72706]

I have not found such a transformer yet, but that would be nice. I'll be using a conservative circuit-breaker/fuse on the primary side, certainly.

I wonder, what could I do to ensure that the control circuitry doesn't drop out under a fault condition? (short circuit leading to voltage droop, for instance)

Thanks for the tip about SCRs; I knew those were useful for something!

EDIT: There's a very strong chance that I'm going to be using a switching pre-regulator, but I haven't given it a ton of thought yet.

[nctnico]

Using a preregulator can be helpfull. About 15 to 20 years ago I designed (*) and built the PSU (2x60V 3A) in the picture below. It has a pre-regulator stage which 'disconnects' the secondary winding when the voltage on the capacitors is 5V above the required output voltage. Because the layout of the PCB isn't that good a lot of the 100Hz switching noise appears on the output. It has the size of a shoe-box but it can deliver a lot of power.

(*) This is the first design I simulated and used MOSFETs for both switching and linear operation.

[digsys]

C'mon F_E, you're a better engineer than all that "bad mouthing" you're shelling out in this thread.
The poor components never did anything wrong to you. In their defense -
The LM317s are excellent regulators. So what, there's 100s much better now (depending on what
you class as "better"), but it doesn't mean everything before is suddenly cr@p. Geez mate.
I've spent parts of the last 2 yrs upgrading a huge inventory of Industrial P/Supplies that are ALL
based on the LM317s, mostly 5-20A. They've sold many 1000s in 30 yrs. As for the myths -
They have built in Short CCT protection, the LM317HV T+K packages up to 60V IN-OUT !!
They have Thermal protection, and (max) current limiting. It is VERY easy to add -
Input protection (1 diode), O/Voltage (TVS and polyfuse), Adjustable current limiting and voltage.
And YES, I've designed with DOZENS of other chips as well, including the LT3080 - ALL have a
purpose in life. PS: The MC34063 was one of my favourite S/Ms years ago, poor thing.
Sheesh, next someone will tell me that the ~100 XP PCs I look after are all KAKA :-)

[David_AVD]

As a new member, I find it curious (and a little disturbing) that a simple "I have these for sale" thread can turn out like this.   

[digsys]

As a new member, I find it curious (and a little disturbing) that a simple "I have these for sale" thread can turn out like this.   

Ah yes, forums are a unique life form of their own :-) You never know WHAT they'll grow in to !!

[free_electron]

C'mon F_E, you're a better engineer than all that "bad mouthing" you're shelling out in this thread.

i'm not bad-moutinhg the chips. i'm bad mouthing 'lab-grade' power supplies built with an lm317 and two resistors. That simply is NOT lab-grade.

Yes LM317 will work perfectly in industrial applications. But they are all pre-set to a specific output voltage and will , during their lifetime , NEVER be set to something else..
You can rip apart any test equipment from HP Agilent keithley and you will find heaps of lm317 and lm337. Nothing wrong, they perform even better than an 78xx , and you only need to keep one part in inventory, you buy them in larger quantity so overal it's cheaper. i fully agree with all that. I too make usage of those parts . no problem there.

But all those applications have one thing in common : the lm317 is set to give a specific output voltage and it will stay there. FOREVER.

None of those designs wil have potentiometers attached on them where someone changes the settings 50 times a day.
None of those designs will have 30 different loads attached a day , some with shorts , some with overvoltages, some in reverse.

They do their work hidden away in some system and the circuitry around them has been tuned for 1 operating point. for all means and purposes they only make 1 output voltage their whole life. it may have been trimmed in the factory or field, but it doesn't change every 2 seconds. The circuitry around them has been optimized for that operating point.

That is a totally different scenario with making a lab power supply. Those things need to be built to take a beating , and stay stable under the most crazy conditions. So , yes , whenever someone exclaims that he is going to make a 'lab' power supply with an LM317 , my neck-hair bristles... and i will bad-mouth that. It is not a proper lab grade power supply. Period ! The LM317 , or any integrated regulator , is not robust enough, nor fulfills the requirements for that purpose. Trying to shoehorn the functionality in with trickery is just plain wrong. It is not meant to be used that way and was not designed for that.

Get a big fat transsitor on a heatsink , make a control loop for voltage and current sensing ( properly done .. and not by using one lm317 as a pre-current limiter to feed another one to do voltage regulation.. your load regulation and impulse response goes to hell with that  approach).
All it really takes is two or three opamps and two transistors. Add some big fat protection diodes ( reverse output , and one across the pass element ) and you will have something that is virturally indestructible.

A lab power supply needs proper current limiting with voltage foldback. Yes, The lm317 will protect itself when you hit 1.5 ampere , or you go over thermal but that is NOT adjustable current limit like you expect from a lab supply. And you'll never be able to burn off enough power in it before it thermally shuts down. no matter how big a heatink you use. the Rthjc of a to220 is not that stellar...  yes you can use the TO-3 version but that is pricey , and still only 1.5 ampere max... you're better off slapping a 2n3055 (or something a bit more modern. there's all these beautiful transistors these days, yet we all seem to be stuck on the 2n3055) in then if you need a to-3 package... more bang for the buck , more power dissipation and more current handling.

The LT3080 is a very nice design. you can put them in parallel and it is intended to replace an lm317 where you need more current (by paralleling) . The voltage control using a steering current is also very clever. It self-balances if you parallel them. It also has a wider working range since they split the regulation supply off the pass element supply. Perfect, great, brilliant , champagne and cigars all around ! i'll happily use it . But NOT as a LAB power supply ! i may use it in a lab supply to make the internal voltages for the opamps to run , but it will not be used as the regulating element , nor as the regulating loop !

The reason i am so adamant about it is that i see a ton of these things coming from wingpangpong. And it invariably dies prematurely. it is not done properly Point two  is that the web is full of 'arduino-style' engineers. Just because they know how to a tie a potmeter to the adj pin makes them believe they have made something that is nobel-prize material... this is just so wrong at so any levels. it is the easy way out. the lazy way out. the wrong way out.. i simply can't work that way. it just doesn't feel right and it goes against proper design. so yeah , when i see another one pop up... i get that itchy tingling feeling, and i just need my soapbox to 'vent'.

Electronics is a passion, i want to do things the right way. At the same time i have no problem explaining how to do things. you can ask me anything , over and over. i will explain until you understand, but i will explain you how to do things the right way. I don't do "slap it togehter and see what happens", that's for junk. If it's worth doing, it's worth doing right. I do not want to be a 'duct-tape and super-glue engineer'.

Me too i want to learn new things. The right things. we are supposed to move forward. not be stuck with 40 year old designs. "Because we've always done it that wy" is NOT the right answer if i have a question. Neither is "because it needs to be cheap" , "because we are lazy" , "because we can't be bothered" or "because we found it on the web , we don't understand it but we blindly copied it and introduced some problems of our own". You can't ever learn something, or move forward that way. you keep mucking about in mud ... you can't move forward if all you use is lm317 and 2n3055... why even bother to design newer parts. if everything can be slapped together with lm317's . let's stop making all those other products and designs. 640k ram ought to be enough for anyone ... right ?

The goal of hobby electronics should be to learn something nad to make things better. Not to blindly solder some parts on a board to make an LED blink. That is not hobby electronics. That's just playing and trampling over a beaten path.

As for the mc34063: i simply don't like that thing. It is too easy to destroy , the regulation loop is old fashioned , it needs big honking inductors and large filtercaps , has no thermal protection , fries itself for no sensible reason ( i've seen some that failed to start properly. they kicked in with the switch full on , without ever turning off..  you need an external current lim resistor.. too much junk needed.

Give me something like an LT1616, LT1930 , that switches in the MHz range, monitors the coil current , can detect coil saturation and turn off and has all kinds of other protections in it. and if possible have synchronous rectification.

 I prefer any of the national semiconductor "Simpleswitchers" over that mc340xx. They have way better performance,  need less stuff around them and are better protected.

right. time to put the soapbox away. It's past midnight here, and tomorrow it's party time (4th of july)